Laser scanning and algorithms tested to make urban trees more wind-resistant

A team of biologists and engineers from the University of São Paulo (USP) has applied LiDAR (light detection and ranging) technology to assess tree health and optimize pruning with the goal of reducing the risk of trees falling in urban areas. The researchers scanned a rosewood tree (Tipuana tipu) on the USP campus, generating a point cloud of 30 million points that reproduced the tree's exact architecture in a computer model. Leaves were digitally removed to analyze the woody structure.

Using finite element method (FEM) simulations, the team modeled wind forces from various directions on the digital tree to identify regions of greatest mechanical vulnerability, or compliance. A pruning algorithm based on topological optimization then calculated which branches should be removed so the tree could redistribute stress and become more balanced. The algorithm aims to remove no more than 20% of the tree's total mass, targeting areas with higher mechanical flexibility to improve wind response.

The study, published in May in the journal Trees: Structure and Function, serves as a proof of concept demonstrating that mathematical equations combined with LiDAR data can guide pruning decisions. Coordinator Marcos Silveira Buckeridge of the Laboratory of Ecological Plant Physiology (LAFIECO) noted that improper pruning often leaves trees vulnerable to wind, a problem worsened by urban canyons that accelerate gusts. In December 2025, winds over 90 km/h caused 1,327 reported fallen-tree incidents in the São Paulo Metropolitan Area, leaving more than 2 million people without power.

The scanned rosewood was located in a wind tunnel with neighboring trees only on its sides, exposing it directly to frontal winds. The researchers found the tree was stronger on the sides where it lacked neighbors, noting that clusters of intertwined trees dissipate wind more effectively than isolated specimens. The methodology applies to eudicotyledonous angiosperms — a large group of flowering plants with classic branching structures — but not to palms, which have a different architecture.

Buckeridge emphasized that the tool is intended to assist human arborists, not replace them. A project in São Paulo is already using LiDAR to scan the city's 650,000 street trees, though at a lower resolution than the single-tree study. Currently, the technique is used for monitoring and cataloging rather than guiding pruning operations. Scanning a single tree at the study's detail level takes about 40 minutes.

The team acknowledges limitations: the current model does not include root systems, which account for roughly 30% of tree falls in São Paulo. A separate submitted study uses ground-penetrating radar to map roots in different urban settings. Other factors not yet modeled include wood expansion and contraction from temperature changes and increased weight from water absorption during prolonged rain, both of which affect fall risk. The researchers have acquired dendrometers to measure these dynamics.

Beyond pruning optimization, the group is using LiDAR for broader tree health diagnostics. On a campus avenue, a vehicle-mounted scanner traveling at 20 km/h maps trees with centimeter-level accuracy, detecting trunk cavities and canopy issues. In another experiment, researchers combined a penetrometer and ultrasound to identify Ganoderma fungus infections in sibipiruna trees, then used LiDAR to compare canopy differences between infected and healthy trees. Biochemical sampling is also underway to search for universal vulnerability markers across species.